Meroplankton distribution and its relationship to coastal mesoscale hydrological structure in the northern Bay of Biscay (NE Atlantic) Sakina-Dorothée Ayata, Robin Stolba, Thierry Comtet, Éric Thiébaut

To cite this version:

Sakina-Dorothée Ayata, Robin Stolba, Thierry Comtet, Éric Thiébaut. Meroplankton distribution and its relationship to coastal mesoscale hydrological structure in the northern Bay of Biscay (NE Atlantic). Journal of Research, Oxford University Press (OUP), 2011, 33 (8), pp.1193-1211. ￿10.1093/plankt/fbr030￿. ￿hal-00854762￿

HAL Id: hal-00854762 https://hal.archives-ouvertes.fr/hal-00854762 Submitted on 28 Aug 2013

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés. 1 Meroplankton distribution and its relationship to coastal mesoscale hydrological structure in 2 the northern Bay of Biscay (NE Atlantic) 3 4 Sakina-Dorothée Ayata a, b, c,*, Robin Stolba a,b, Thierry Comtet a,b, Éric Thiébaut a,b 5 6 a Université Pierre & Marie Curie, Station Biologique de Roscoff, Laboratoire 'Adaptation et 7 diversité en milieu marin', Place Georges Teissier, F-29680 Roscoff cedex, France. 8 b CNRS, UMR 7144, F-29680, Roscoff cedex, France. 9 c Present address: Laboratoire d’Océanographie et du Climat: Expérimentation et Approches 10 Numériques - Institut Pierre Simon Laplace, UMR 7159, BC 100, 4 Place Jussieu, F-75252 Paris 11 cedex 05, France. 12 13 * Corresponding author: [email protected] 14 15 Keywords: Meroplankton, larval transport, mesoscale structures, river plume, fronts, Owenia 16 fusiformis, Pectinaria koreni, Sabellaria alveolata, Bay of Biscay. 17 18 19 ABSTRACT 20 The relationship between meroplankton distribution and spatio-temporal variability of coastal 21 mesoscale hydrological structure was investigated in the northern Bay of Biscay, North-East 22 Atlantic. For the three coastal polychaetes studied, i.e., Pectinaria koreni, Owenia fusiformis, 23 Sabellaria alveolata, the highest larval abundances were sampled in low salinity, low density and 24 high temperature river plume waters. For two species (P. koreni and O. fusiformis), maximal 25 abundances were observed in the surface and thermocline layers due to ontogenic migrations. 26 Variance partitioning based on multiple regression and redundancy analyses were used to assess the 27 relative roles played by the hydrological environment alone, the geographical space alone, and their 28 interactions, i.e. the spatial structure of the hydrological environment. These analyses demonstrate 29 the key role played by hydrological spatial structure in the distribution of larval abundances, 30 indicating that larvae mainly act as passive particles. The hydrological environment alone was 31 insignificant, whereas geographical space alone explained a significant part of the variability in 32 meroplankton distribution, probably in conjunction with ecological processes. For species whose 33 benthic populations are spatially structured, the distribution and the size of adult populations and 34 the timing of spawning events can significantly affect larval distribution and dispersal.

1 1 INTRODUCTION 2 For marine benthic with a complex life cycle, pelagic larval transport is a key process 3 in dispersal and population connectivity and plays a major role in population establishment and 4 persistence, biodiversity conservation, spread of alien species and species distributions (Cowen and 5 Sponaugle, 2009). Larval transport results from complex interactions of biological traits (e.g., 6 spawning time and location, larval behaviour, planktonic larval duration) and hydrodynamic 7 processes, and both vary at several spatial and temporal scales (Pineda et al., 2007). In coastal 8 environments, numerous and complex mesoscale hydrodynamic features can tightly control the 9 transport of planktonic organisms (Largier, 2003). For example, coastal larvae may be 10 “trapped” and transported within estuarine plumes, which act as physical barriers to offshore 11 transport (Thiébaut, 1996; Shanks et al., 2002; Shanks et al., 2003a). Biological species-specific 12 traits may also modulate the role of this type of hydrodynamic feature on larval transport. In 13 particular, the vertical behaviour of some larvae may ensure their retention in suitable habitats, even 14 when local hydrodynamics is expected to induce offshore transport, as in estuaries or 15 areas (Thiébaut et al., 1992; Shanks et al., 2003b; Queiroga et al., 2007). In addition, spawning 16 time and location may be more important than larval behaviour in determining larval dispersal 17 (Edwards et al., 2007; Ayata et al., 2010; Carson et al., 2010). Sorting out which factors control 18 larval distribution and transport thus remains a challenging issue for better understanding marine 19 population connectivity and metapopulation dynamics (Gawarkiewicz et al., 2007; Pineda et al., 20 2007). Although new technological advances (e.g. biophysical models, molecular and geochemical 21 markers) and the integration of different disciplines has fostered recent progress in assessing larval 22 dispersal (Levin, 2006), mesoscale surveys are still needed to explicitly evaluate the role of most 23 physical processes on larval populations. 24 In the Bay of Biscay (North-East Atlantic, Fig. 1), numerous mesoscale hydrological 25 features, including frontal systems, river plumes, wind-induced coastal or low salinity 26 lenses, have been reported to occur simultaneously and to have large spatio-temporal variability at 27 scales ranging from days to seasons (Koutsikopoulos and Le Cann, 1996; Puillat et al., 2006). 28 These features have been demonstrated to play a role on primary production (Morin 29 et al., 1991; Maguer et al., 2009), community structure (Albaina and Irigoien, 2007), 30 and plankton biomass distribution (Zarauz et al., 2007; 2008). Furthermore, the northern Bay of 31 Biscay forms a transition area between the Boreal-Arctic and the Boreal-Lusitanian biogeographic 32 provinces together with the occurrence of phylogeographic breaks in several marine taxa (Jolly et 33 al., 2006; Maggs et al., 2008). Through their influence on larval transport and population 34 connectivity, the contemporary complex hydrodynamics may participate in the maintenance of 35 these genetic breaks (Ayata et al., 2010).

2 1 In this general context, we focused on the horizontal and vertical larval distributions in three 2 target species of coastal polychaetes for which complex phylogeographic patterns have been 3 reported in the northern Bay of Biscay: Pectinaria koreni, Owenia fusiformis and Sabellaria 4 alveolata (Jolly et al., 2006; F. Rigal and F. Viard, Station Biologique de Roscoff, unpublished 5 data). For these species, pelagic larvae can be morphologically identified at the species level and the 6 extended spawning seasons increase the probability of successfully sampling larvae: (1) a major 7 spawning period in spring (April-June) and an additional spawning period in late summer-autumn 8 for P. koreni (Irlinger et al., 1991); (2) a main spawning period in spring (April-June) for O. 9 fusiformis (Gentil et al., 1990) and (3) year-round reproduction with two main reproductive peaks 10 in March-April and June-July for S. alveolata (Gruet and Lassus, 1983; Dubois et al., 2007). 11 Although these three species share the same general broadcast spawner life history, they differ in 12 their adult benthic habitats, planktonic larval durations and larval behaviours, and all of these 13 biological traits are known to significantly affect larval dispersal. P. koreni and O. fusiformis 14 inhabit patches of fine sand and muddy fine sand in shallow coastal waters at about 10-20 m depth 15 (Chassé and Glémarec, 1976; Fig. 1) whereas S. alveolata is a gregarious polychaete that builds 16 intertidal biogenic reefs. Large S. alveolata reefs have been described on sandflats of the Bay of 17 Bourgneuf south of the Loire estuary (Gruet and Lassus, 1983), while small groups of individuals 18 adhering to rocks have also been reported all along southern Brittany (F. Rigal and F. Viard, 19 Station Biologique de Roscoff, unpublished data) (Fig. 1). From in situ observations, mean 20 planktonic larval durations have been estimated at about two weeks for P. koreni (Lagadeuc and 21 Retière, 1993), four weeks for O. fusiformis (Thiébaut et al., 1992), and four to ten weeks for S. 22 alveolata (Dubois et al., 2007). Moreover, these species differ in their potential larval behaviour: 23 vertical ontogenic migrations have been reported for P. koreni (Lagadeuc, 1992) and O. fusiformis 24 (Thiébaut et al., 1992), whereas complex short-term variations in S. alveolata vertical distribution 25 have been reported in relation to the tidal cycle (Dubois et al., 2007). Focusing on these three target 26 species, the aim of the present study was to assess the relative role of coastal mesoscale 27 hydrological structure and biological processes in meroplankton distribution, as well as larval 28 dispersal and connectivity. 29 Several statistical methods that model spatial and temporal relationships in ecological data 30 have been recently developed to describe the distribution of zooplankton in relation to their 31 environment (e.g., generalized linear models, generalized additive models, clustering or ordination) 32 (e.g. Zarauz et al., 2008), but these methods neglect the spatial structure of the environmental 33 variables. However, the spatial patterns of a response variable such as larval distribution depend on 34 (1) the hydrological environment alone (i.e. non-spatial environmental variation); (2) the 35 geographical space alone (i.e. spatial variations not shared by the environmental variables); (3) the

3 1 interactions between space and environment, i.e., the spatial structure of the hydrological 2 environment and (4) other undetermined factors such as organism behaviour (Borcard et al., 1992). 3 Hence, the spatial structure of the environment should be taken into account when describing spatial 4 patterns of zooplankton (Belgrano et al., 1995), for example by using variance partitioning, a 5 recently developed quantitative statistical method. 6 The application of a variance partitioning method to mesoscale samples of larval horizontal 7 distribution in two different oceanographic conditions provided the opportunity to test a set of 8 hypotheses on the main factors affecting the meroplankton distributions and larval dispersal in the 9 northern Bay of Biscay and to quantify their relative roles. Under our first hypothesis, larval 10 distributions would mainly depend on the hydrological properties of the water masses (e.g. salinity, 11 temperature, stratification of the water column) that may affect larval development and survival. 12 According to the second hypothesis, it is assumed that larvae are advected passively by currents and 13 trapped by mesoscale hydrodynamic structures within typical water masses. Consequently, 14 spatially-structured hydrodynamic processes would be the major factor affecting the meroplankton 15 distribution and similar larval distributions would be observed for the different larval stages of 16 target species and for the different target species. Finally, under the third hypothesis, biological 17 traits such as spawning, settlement location and swimming larval behaviour would be the major 18 factors determining larval distributions and significant differences would be observed between the 19 species characterised by different biological traits. To test this last hypothesis, variance partitioning 20 was completed by an analysis of larval vertical distribution in relation to water column stratification 21 in a few stations where larval densities were high. 22 23 METHODS 24 Study area 25 In coastal areas of the Bay of Biscay, strong freshwater inputs combined with relatively low vertical 26 mixing cause strong haline stratification. Thermal stratification superimposes on haline 27 stratification, not necessarily at the same depth, from spring to mid-September. Vertical temperature 28 gradients reach a maximum of 9 to 10°C between surface and bottom layers. In the vicinity of the 29 Loire and Vilaine estuaries (Fig. 1), the presence of low salinity surface waters in spring induces 30 significant density gradients responsible for strong density currents over the continental shelf (2 to 31 20 cm s-1) generally oriented northwards. The wind-induced currents are highly variable both in 32 direction and speed at temporal scales ranging from days to seasons, and typically reach 10 to 30 33 cm s-1 in the north of the bay. During thermal stratification, W to NW winds induce local transitory 34 upwellings in the north of the Loire estuary and result in the formation and offshore transport of 35 lenses of low salinity surface waters about 50-80 km wide and 30 m thick (Puillat et al., 2006). In

4 1 the north, the Bay of Biscay is connected to the English Channel at the tip of Brittany through the 2 Ushant Sea, a transitional area between well-mixed (English Channel) and stratified (Bay of 3 Biscay) waters where strong thermal fronts are caused by tidal mixing in spring and summer 4 (Pingree et al., 1975; Morin et al., 1991). Low salinity waters from the Loire and Vilaine Rivers 5 have been reported from March to April in the Ushant Sea and at the entrance of the English 6 Channel following strong river run-offs and periods of NE winds (Kelly-Gerreyn et al., 2006). 7 8 Sampling and laboratory procedures 9 Data were collected in the northern Bay of Biscay in spring 2008 during two oceanographic cruises 10 (i.e. from 10 to 18 May and from 9 to 13 June). Cruises were separated by an interval of 3 weeks to 11 increase the probability of sampling larvae of the three target species and to cover different 12 oceanographic conditions during the main reproductive season. Along seven inshore-offshore 13 transects localized in front of the main coastal bays and estuaries in the study area, eight stations 14 distributed from the 20 m isobath to the open sea were sampled during the daytime (Fig. 1). The 15 first three stations were separated by 3 nautical miles and then four stations 6 nautical miles apart, 16 so that transects extended about 40 nautical miles offshore. 17 At each sampling station, vertical profiles of hydrological parameters were obtained from 18 the surface to the bottom (or to a maximum of 50 m depth) using a CTD probe (Seabird SBE 19). A 19 stratification index ! of the water column was defined by integrating the vertical density profile as 20 the mean of the density differences along the water column (Fortier and Leggett, 1982): i=n 21 ! = (!i=1 "#ti . "zi)/n

22 where n is the number of pairs of adjacent measurements, "#ti the difference in water density th th 23 between the i pair of measurements and "zi the depth interval between the i pair of measurements 24 fixed at 1 m in the present case. 25 The thermocline, halocline, and pycnocline depths were calculated using a theoretical two-layer 26 model of the water column (Planque et al., 2006). For a given hydrological variable X, the water

27 column is assumed to be composed of two homogeneous layers, a surface layer of width zx, and a

28 bottom layer of width zb-zx. The cline depth zx is

29 zx = zb |Xm-Xb| / |Xs-Xb|

30 where zb is the water column height, Xm the mean value of variable X from surface to bottom, Xb the

31 value of variable X in the bottom layer, and Xs the value of variable X in the surface layer. The

32 variable X stands for temperature T, salinity S, or density #t to calculate the thermocline, halocline, 33 or pycnocline depth, respectively. 34 In addition to in situ data, satellite images of sea surface chlorophyll a concentrations during 35 the two cruises were obtained from CERSAT (French Centre for European Remote-Sensing

5 1 Satellite Processing and Archiving Facility). Wind data in spring 2008 were provided by the French 2 Office of Meteorology (Météo France) and were recorded at two locations in the north (i.e. 3 Penmarc’h headland) and in the centre of the study area (i.e. Belle-Ile Island) (Fig. 1). Since wind 4 data recorded at the two semaphores were very close, indicating homogeneous wind conditions all 5 over the study area, only data collected at Penmarc’h will be presented. Daily records of the Vilaine 6 and Loire River flows were provided by the French Freshwater Office database. 7 To describe the horizontal larval distribution, zooplankton samples were collected at each station 8 using a vertical haul from the bottom (or a maximum depth of 50 m) to the surface with a triple 9 WP2 plankton net with a mesh size of 80 !m and a mouth area of 0.25 m2. A TSK flow meter 10 (Tsurumi Seiki Co., Ltd., Yokohama, Japan) was fixed to one of the net apertures to determine the 11 volume of filtered water (range: 2.6-19.9 m3, average: 10.0 m3). The samples collected by one of 12 the three nets were preserved in 3% buffered formalin and analysed. 13 To determine the larval vertical distribution in relation to water column stratification, 14 zooplankton samples were also collected at fixed depths using a pumping system. A 5 cm diameter 15 collection pipe was attached to the CTD armature and to an immerged pump. CTD records provided 16 real-time data on the depth and the water properties from which samples were collected. Three to 17 four depths were selected per station: the surface mixed-layer (about 3-5 m), the halocline and/or 18 thermocline layers (about 10-15 m), and the bottom layer (about 20-40 m). With a pumping rate of 19 300 L min-1, a volume of 1.5 m3 of water was filtered through an 80 !m mesh net at each sampled 20 depth for 5 min. Samples were preserved immediately after collection in 3% buffered formalin. 21 Owing to the time required to sort samples, larval vertical distributions were described for only a 22 few stations with high larval abundances (i.e., higher than a few hundreds of ind.m-3) (i.e., stations 23 D2, C1, P2, and V2 in May and stations D2, P1, and V1 in June, see Fig. 1). 24 In the laboratory, each sample was rinsed and completed to 200 mL with filtered seawater. 25 Samples were then homogenized by vigorous random stirring and a subsample of 5 mL was 26 pipetted. The larvae of the three target species were enumerated in 3 to 5 subsamples under a 27 dissecting stereomicroscope, to reach a count of at least 100 larvae per sample and species 28 (Frontier, 1972). When total zooplankton concentration was low, the sample was completed to 100 29 mL only. Conversely, when concentration was high, the subsample volume was decreased to 3 mL. 30 By this procedure, 3% to 20% of each sample was analysed. Larval abundances were expressed as 31 number of larvae per m3. For each species, the four larval developmental stages were distinguished 32 using morphological characters (P. koreni: Lagadeuc and Retière, 1993; O. fusiformis: Wilson, 33 1932; S. alveolata: Wilson, 1968; Dubois et al, 2007). 34 35 Statistical analyses

6 1 All statistical analyses were performed using R software, version 2.7 (R Development Core Team, 2 2005; http://www.r-project.org) and the vegan library. 3 For each cruise, the hydrological typology of the water masses was based on ten variables:

4 the sea surface temperature (Ts), the temperature gradient from surface to bottom or a 50 m depth

5 ("T), the thermocline depth (zT), the sea surface salinity (SS), the salinity gradient from surface to

6 bottom or at 50 m ("S), the halocline depth (zS), the surface density (!tS), the density gradient from ! 7 surface to bottom or at 50 m ("!t), the pycnocline depth (z!t), and the stratification index ( ). A 8 cluster analysis was carried out using Ward's method on the Euclidean distance matrix calculated 9 between stations to group stations according to their physico-chemical properties and to identify 10 hydrological regions. 11 The horizontal depth-integrated larval abundances of each species were analysed using a 12 variance partitioning method based on three partial multiple regressions (Borcard et al., 1992; 13 Legendre and Legendre, 1998). For these analyses, three vectors or matrices were used: (1) the

14 response vector Y of the log10-transformed horizontal larval abundances; (2) the explanatory matrix 15 X of the hydrological variables and (3) the explanatory matrix W of the spatial variables. To ensure 16 the extraction of the more complex structures, spatial variables were described as the different terms 17 of a cubic polynomial function of the centred geographical coordinates (x,y) of the stations such that 18 W = f(x,y,x2,xy,y2,x3,xy2,x2y,y3). In the first partial multiple regression step, larval abundances were 19 analysed as a function of the hydrological data, through a multiple regression between the response 20 vector Y and the environmental matrix X. Since multiple regression is sensitive to colinearity among 21 the explanatory variables, variables were selected prior to the regression to avoid redundancy 22 between the ten hydrological variables. The matrix of Pearson's product-moment correlation 23 coefficients was calculated between each pair of hydrological variables and only variables with a 24 correlation coefficient lower than 0.75 were chosen for the analysis so that one variable was omitted 25 for each pair of highly correlated variables. Hence, four hydrological variables covering the spatial 26 variations of salinity and temperature and the vertical structure of the water column were selected in ! 27 May (Ts, SS, zS, z!t) and in June (Ts, SS, z!t, ). In the second step, a multiple regression was 28 performed between Y and the spatial matrix W. Finally, in the third step, a multiple regression was 29 performed between Y and both the hydrological and spatial matrix XW. From these three successive 30 steps, variance in total larval abundances was partitioned: the first regression provided the 31 proportion of variance explained by the hydrological environment (a+b: hydrological environment 32 alone and interaction between hydrological environment and geographical space), while the second 33 regression indicated the proportion of variance explained by the geographical space (c+b: 34 geographical space alone and interaction between hydrological environment and geographical 35 space). The third regression provided the proportion of variance explained by the hydrological

7 1 environment, the geographical space, and their interaction (a+c+b) and allows the calculation of the 2 unexplained variance (d = 1-a-b-c). 3 To analyse the horizontal distributions of the different larval stages, redundancy analyses 4 (RDA), which are a direct extension of multiple regressions to multivariate response data, were 5 performed and variance partitioning was conducted as previously described (Legendre and 6 Legendre, 1998). 7 8 RESULTS 9 Meteorological and run-offs conditions in spring 2008 10 Wind conditions were mainly from W to NW during spring 2008, with decreasing average speed 11 over time, from 8.5 m.s-1 in March to 5.6 m.s-1 in April and 3.3 m.s-1 in May (Fig. S1A). In June, 12 average wind speed increased slightly to 4.8 m.s-1. The May and June cruises occurred under low to 13 moderate wind speeds of 2.6 and 6.0 m.s-1, respectively. After the May cruise, low winds of various 14 directions (average speed, 3.9 m.s-1) were recorded over ten days. The week before the second 15 cruise was characterized by stronger NW winds (average speed, 6.4 m.s-1), with maximal speed of 8 16 to 9 m.s-1. Two peaks in Loire River run-offs were observed in March and April 2008, with run-offs 17 exceeding 1,800 and 2,500 m3.s-1, respectively (Fig. S1B). The May cruise occurred during a period 18 of low river run-offs, about 3 weeks after the last flood. Just before the June cruise, a peak in the 19 river outputs of both the Vilaine and the Loire Rivers was recorded (> 2,000 m3.s-1 for the Loire). 20 21 Environmental variables during the May cruise

22 In May 2008, a large low salinity plume (SS < 32) was observed along the southern Brittany coasts 23 from the Loire estuary to the entrance of the Bay of Concarneau (Fig. 2A). Surface salinities were

24 minimal at the most inshore stations of the Loire transect (SS = 26.5 at station L1) and maximal 25 along the two transects of Douarnenez and Audierne and at the more offshore stations of the other

26 transects (SS > 34). Sea surface temperatures followed a latitudinal gradient with the lowest

27 temperatures in the north (TS < 13.5°C at stations D8 and D7) and the highest temperatures in the

28 south (TS > 16.7°C at stations L2 and L3) (Fig. 2B). Cross-shore thermal gradients were weak, 29 except along the Douarnenez transect where a thermal front was observed at the entrance of the bay 30 and along the Loire transect. The highest chlorophyll a concentrations were localized in river plume 31 waters (Fig. 2C). Vertical stratification was low (! < 0.1), except for the coastal stations located in 32 the river plumes (Fig. S2). Weak thermal stratification was observed along most transects. When 33 observed, the depths of the thermocline and the halocline differed, and varied between 5 and 10 m 34 and between 10 and 15 m, respectively. The distribution of water density was mainly governed by 35 variation in salinity as illustrated by the similarity of the salinity and density profiles (Fig. S2B,

8 1 S2C). 2 The cluster analysis distinguished two main regions according to their hydrological 3 properties (Fig. 3A-B). The first cluster grouped the inshore stations of the southern transects, 4 located in river plumes and characterised by higher stratification, lower surface salinities and 5 densities and shallower thermoclines, haloclines and pycnoclines (Table I). The second cluster was 6 composed of the stations along the northern transects and of the offshore stations along the southern 7 transects. It could be further separated in two subgroups that differed by their surface temperatures 8 (Table I). 9 10 Horizontal larval distribution during the May cruise 11 Inshore-offshore and latitudinal gradients were observed in larval distributions of all three species, 12 with higher abundances (> 100 ind. m-3) close to the shore and in the southern part of the study area 13 (Fig. 4). P. koreni and O. fusiformis larvae were sampled along each transect, but S. alveolata 14 larvae were only observed along the five southernmost transects, from the Bay of Concarneau to the 15 Loire estuary. The P. koreni larval population was mainly composed of the first three larval 16 developmental stages (Fig. 5A), which were homogeneously distributed over the study area (data 17 not shown). In contrast, the fourth aulophore stage was only reported in a few stations. The larval 18 population of O. fusiformis was largely dominated by stage-3 larvae and to a lesser extent by stage- 19 4 larvae (Fig. 5B). While these two larval stages were evenly distributed, the two youngest stages 20 were restricted to the more coastal stations. The S. alveolata larval population was also largely 21 dominated by a single larval developmental stage: stage 2 (Fig. 5C). 22 For all three species, multiple regressions showed significant relationships between total 23 larval abundances and the four selected hydrological variables: sea surface temperature, sea surface 24 salinity, halocline depth and pycnocline depth (P. koreni: R2 = 0.5669, p < 10-3; O. fusiformis: R2 = 25 0.4486, p < 10-3; S. alveolata: R2 = 0.6960, p < 10-3). Larval abundances were significantly 26 negatively correlated with surface salinity and either surface temperature or halocline depth, 27 indicating that larvae were mainly located in the river plume stations (Table II). Variance 28 partitioning highlighted that variation in larval abundances was mainly explained by the spatial 29 structure of the hydrological environment (56% for P. koreni, 44% for O. fusiformis and 68% for S. 30 alveolata) (Fig. 6A). For all three species, geographical space alone accounted for 26%, 27% and 31 16% of the variance, respectively, whereas the non-spatial hydrological environment explained less 32 than 1% of the variance. The remaining unexplained variance ranged between 14% and 28%. 33 The RDA computed on the abundances of the different larval stages and the hydrological 34 variables indicated that the hydrological variables significantly influenced larval stage distribution 35 (P. koreni: R2 = 0.5278, p < 10-3; O. fusiformis: R2 = 0.3072, p < 10-3; S. alveolata: R2 = 0.5582, p <

9 1 10-3). For example, for the RDA performed on P. koreni larvae, only one canonical axis, which 2 explained 94.2% of the constrained variance, was significant (p<0.001) and was related to the 3 surface salinity and the pycnocline/halocline depths (Fig. 7A). The scores of the first three 4 developmental stages on the RDA biplot diagram were grouped together, suggesting similar 5 horizontal distributions according to the hydrological environment in the plume waters. Only the 6 score of stage 4 differed slightly, which may be partly explained by its very low occurrence. 7 Variance partitioning from the successive redundancy analyses showed that the spatial structure of 8 the hydrological environment accounted for 52%, 30% and 55% of the variation in larval stage 9 distributions for P. koreni, O. fusiformis and S. alveolata, respectively (Fig. 6B). Geographical 10 space alone explained respectively 22%, 25% and 22% of the variance in stage abundances, 11 whereas the hydrological environment alone explained less than 1% of the variance. The 12 unexplained variance varied between 22% and 44%. 13 14 Environmental variables during the June cruise 15 Similar to what was observed in May, surface waters showed the lowest salinities off the Vilaine

16 and Loire estuaries (SS < 32; Fig. 2D). However, although the river plume was less salty (minimal

17 salinity: 29.8), a northern and offshore extension of low salinity waters was observed (Ss < 34).

18 Surface waters were generally warmer than in May (TS > 14.5°C), except at stations D7 and D8 19 separated from the other stations of the Douarnenez transect by the Ushant thermal front (Fig. 2E). 20 Maximal sea surface temperatures were observed offshore from the Bay of Concarneau to the Loire

21 estuary (TS > 17.9°C). Chlorophyll a concentrations were high all along the continental shelf with 22 maximal values in inshore waters in response to the spring phytoplankton bloom (Fig. 2F). 23 Compared to data obtained in May, a stronger vertical stratification was observed (0.03 < ! < 0.19) 24 except at the offshore stations of the Douarnenez transect (Fig. S3). Along the Vilaine and Loire 25 transects, spatial variation in water density varied primarily with salinity, whereas spatial variation 26 was mainly related to temperature along the other transects. 27 The cluster analysis again separated two main hydrological regions (Fig. 3C-D). The 28 stations of the first cluster were located off the Vilaine and Loire estuaries, and were characterised 29 by stronger vertical stratification, lower surface salinities, lower surface densities and shallower 30 thermoclines, haloclines and pycnoclines (Table I). The second cluster, which included all the other 31 stations, could be divided in two subclusters according to surface temperature (Fig. 3C-D, Table I). 32 The first subcluster characterized by higher surface temperatures and deeper thermoclines was 33 composed of most of the stations in the transects from Audierne to Etel-Quiberon. Conversely, the 34 second subcluster, characterized by lower surface temperatures and shallower thermoclines, 35 included all the stations of the Douarnenez transect and some inshore stations of other transects.

10 1 2 Horizontal larval distribution during the cruise of June 3 The general horizontal larval distributions observed in June for all three species, resembled those 4 described in May: the percentages of larval occurrence were similar and higher larval abundances 5 were recorded in coastal and/or southern stations (Fig. 4). However, mean larval abundances were 6 lower in June than in May, especially for O. fusiformis and S. alveolata. Other slight differences 7 were observed for P. koreni and O. fusiformis. For P. koreni larvae, a more offshore larval 8 distribution was observed at two southern transects (i.e. Etel-Quiberon and Loire) while higher 9 densities were sampled in the Bay of Douarnenez. The distribution of O. fusiformis larvae was more 10 variable between transects in June than in May, with larvae mainly sampled along the northernmost 11 transect (Douarnenez), and the three southernmost transects. The P. koreni larval population was 12 composed of older stages in June (mainly larvae of stage 3) than in May, while the larval 13 populations of O. fusiformis and S. alveolata were composed of younger stages than in May (Fig. 14 5). Given the planktonic larval duration of the three target species and the temporal variation in the 15 relative proportions of the different developmental stages, it seems unlikely that larvae sampled in 16 June originated from the same spawning events as those sampled in May. For O. fusiformis and S. 17 alveolata, younger larvae were collected in June, indicating that at least one spawning event 18 occurred between the two cruises. For P. koreni, although older larvae were observed in June, they 19 would have originated from a spawning event occurring at the end of May, assuming a planktonic 20 larval duration of about two weeks. 21 Multiple regressions indicated significant relationships between larval concentrations and 22 the four hydrological variables selected in June (sea surface temperature, sea surface salinity, 23 pycnocline depth, stratification index) (P. koreni: R2 = 0.6031, p < 10-3; O. fusiformis: R2 = 0.4513, 24 p < 10-3; S. alveolata: R2 = 0.7212, p < 10-3) (Table III). Significant negative correlations were 25 observed between larval abundances and surface temperature for all three species, but other 26 significant correlations varied among species. Positive significant correlations were observed 27 between larval abundances and stratification index for P. koreni and S. alveolata, and between 28 larval abundances and surface salinity for P. koreni. Conversely, negative significant correlations 29 were observed between larval abundances and pycnocline depth for P. koreni and O. fusiformis. 30 However, as observed in May, variance partitioning indicated that larval horizontal distributions 31 were mainly explained by the spatial structure of the hydrological environment, which accounted 32 for 58% of the variability in total abundances for P. koreni, 40% for O. fusiformis and 60% for S. 33 alveolata (Fig. 6C). For these species, geographical space alone explained 8% to 20% of the 34 variation in the larval abundances, whereas the hydrological environment alone only explained from 35 3% to 12% of their variation. Unexplained variance varied from 15% to 47%.

11 1 For all three species, the RDA indicated significant relationships between larval stage 2 abundances and hydrological variables (P. koreni: R2 = 0.4624, p < 10-3; O. fusiformis: R2 = 0.3507, 3 p < 10-3; S. alveolata: R2 = 0.5383, p < 10-3). For each species, the RDA biplot diagram showed that 4 the first canonical axis, which was the only significant axis (p<0.001), was positively correlated to 5 surface salinity, surface temperature and pycnocline depth, and negatively correlated to the 6 stratification index as illustrated, for instance, for P. koreni larvae (Fig. 7B). The scores of the 7 different larval stages were grouped together on the biplot diagram, indicating that they were all 8 located preferentially in stations with higher stratification and lower surface salinity and 9 temperature. Variance partitioning indicated that variation in larval stage distributions were mainly 10 due to the spatial structure of the hydrological environment which accounted for 39% of the total 11 variation for P. koreni, 28% for O. fusiformis and 39% for S. alveolata (Fig. 6D). For these three 12 species, geographical space alone explained 18%, 7% and 19% of the variations in larval stage 13 abundances respectively. The fraction of the variance explained by the non-spatial hydrological 14 environment reached 7%, 7% and 14%, respectively, and the fraction of unexplained variance 15 varied from 27% to 58%. Hence, the importance of the spatial structure of the hydrological 16 environment in explaining the variations in the distribution of the different larval stages decreased 17 in June, whereas the fraction due to the hydrological environment and the proportion of unexplained 18 variance both increased. 19 20 Larval vertical distributions 21 Average vertical distributions indicated that young larvae (stages 1 and 2) of P. koreni and O. 22 fusiformis were mainly restricted to the surface and halocline/thermocline layers (Fig. 8A-B), 23 suggesting a preferential distribution within river plume waters. For these two species, ontogenic 24 vertical migration was also observed, with a deeper distribution of the oldest larvae (stage 4). In 25 contrast, S. alveolata larvae were not restricted to surface river plume waters but tended to be 26 evenly distributed over the whole water column (Fig. 8C). 27 28 DISCUSSION 29 Meroplankton distribution in relation to hydrological structure 30 Previous descriptions of the mesoscale hydrodynamic features of the Bay of Biscay have 31 highlighted the importance of short temporal variation at scales ranging from a few days to one 32 week (Puillat et al., 2006), which may preclude obtaining a synoptic view of the hydrological 33 conditions from oceanographic cruises (Planque et al., 2006). However, although sampling was 34 performed over five to nine consecutive days in the present study, satellite data indicated that at 35 least sea surface temperatures were similar at the beginning and the end of each cruise (Fig. S4A-C

12 1 for May cruise and Fig. S4D-F for June cruise), indicating that our data provided a synoptic view of 2 the hydrological environment for each cruise. In addition, no abrupt change in wind direction or 3 intensity was reported during sampling, further suggesting that the spatial distribution of 4 hydrological characteristics was not biased by short-term temporal changes. Two of the six 5 hydrological regions previously described by Planque et al. (2006) in spring in the Bay of Biscay 6 were then identified both in May and June 2008, i.e. oceanic waters and river plume waters, 7 although their location and hydrological properties changed sharply between the two cruises with 8 consequences on larval distribution. 9 The hydrological environment may influence larval distribution through two components, 10 the hydrological environment itself and/or its spatial organisation. The spatial structure of 11 hydrological variables can lead to the overestimation of a strictly hydrological role due to high 12 spatial autocorrelation (Legendre and Trousselier, 1988). However, the use of variance partitioning 13 avoids this problem by isolating (1) the non-spatial component of the hydrological environment, (2) 14 the spatially structured component of the hydrological environment, and (3) the non-hydrological 15 spatial component of the total variance in species distribution or community structure (Borcard et 16 al., 1992). By using this approach, our results reveal that the hydrological environment alone 17 explains only a small portion of the total variance in larval abundances, hence refuting our first 18 hypothesis, even though parameters such as salinity or temperature are known to directly influence 19 larval mortality or development (Anger et al., 1998; O'Connor et al., 2007). In contrast, and in 20 accordance with our second hypothesis, the interactions between hydrological environment and 21 geographical space generally explained most of the variation in larval distribution for the three 22 polychaete species (i.e. from 40% to 68% of the total variance). A similar result has been reported 23 for the meroplankton distribution along the coasts of the , a region that is also 24 characterised by the presence of a front separating coastal and offshore waters (Belgrano et al. 25 1995). 26 In our study, the maximal larval abundances of all three coastal polychaete species were 27 mostly sampled in river plume waters, thus confirming their important role in concentrating and 28 transporting coastal invertebrate larvae as reported in various nearshore environments (Thiébaut, 29 1996; Shanks et al., 2002). The spatial patterns in the horizontal larval distribution of each species 30 were explained by the same set of spatially organised hydrological variables despite different 31 spawning locations, planktonic larval durations and larval behaviour, highlighting the dominant role 32 of spatially structured hydrodynamic processes over specific biological traits. Furthermore, while a 33 spatial gradient in larval abundance according to developmental stage is expected in a homogeneous 34 environment — with older larvae located offshore and younger larvae in the vicinity of their 35 spawning location (i.e., close to the ) — no clear pattern of spatial distribution of the different

13 1 larval development stages was observed here, suggesting that larvae of various ages are mixed 2 within the river plume waters. River plumes and associated fronts may act as physical barriers for 3 the dispersal of pelagic larvae, retaining them in the vicinity of their spawning location, or 4 transporting them alongshore, thus limiting their offshore export and allowing connectivity between 5 neighbouring populations (Thiébaut, 1996; Largier, 2003). In addition to this role on larval 6 transport, river plume waters in the Bay of Biscay have been shown to sustain a phytoplankton 7 biomass — predominantly composed of large phytoplankton cells — higher than in offshore waters 8 (Morin et al., 1991; Maguer et al., 2009). This phytoplankton biomass may thus provide 9 planktotrophic polychaete larvae with abundant food resources. 10 The importance of the river plume fronts as physical boundaries will also depend on their 11 temporal variability with regard to planktonic larval duration. As an example, transitory wind 12 events are likely to disrupt the plume frontal systems and induce large-scale dispersal in response to 13 upwelling events (Thiébaut, 1996; Shanks et al., 2003b). However, in areas characterized by a 14 strong seasonal and intra-seasonal variation in hydrodynamic conditions, the timing of the spawning 15 events in relation to the establishment of mesoscale structure is crucial for meroplankton 16 distribution and a change in reproductive phenology would significantly alter larval transport, 17 connectivity and metapopulation dynamics (Ayata et al., 2010; Carson et al., 2010). In the present 18 study, the extent of the river plumes differed between the two sampling dates with consequences on 19 larval distribution. Larvae were retained close to the shoreline in May due to the limited spatial 20 extension of river plumes. Given alongshore density currents of about 10 cm.s-1 along the southern 21 coasts of Brittany (Lazure and Jégou, 1998), an alongshore average advective transport of larvae of 22 about 8.64 km d-1 is likely. However, this rough estimate can be largely influenced by the spatial 23 variability in alongshore currents due to the local topography and the short-term effects of wind- 24 induced currents. Such hydrological structuring can limits cross-shore export of larvae but may 25 promote larval connectivity between neighbouring populations along the coasts of southern 26 Brittany. Conversely, in June, strong N to NW winds recorded the week before the sampling survey 27 could have favoured coastal upwelling and offshore export of plume waters. Although larvae 28 remained mainly confined within plume waters, larval retention close to adult populations, and thus 29 larval connectivity among neighbouring populations, were lower in June than in May, which may 30 greatly affect larval settlement potential. On the other hand, variation in the hydrological 31 environment can alter the relative role of river plumes in controlling larval distribution: the 32 variation attributed to the interactions between hydrological environment and geographical space 33 was lower in June than in May, for the total larval distribution of O. fusiformis and S. alveolata and 34 larval stage distribution of all three species, compared to the temporal variation in the degree of the 35 spatial organisation of the hydrological environment. Accordingly, the fraction of unexplained

14 1 variance was higher in June, i.e. when hydrological structuring was less contrasted due to the 2 dilution of plume waters. 3 4 The relative role of specific biological traits 5 In addition to the spatial organisation of the hydrological environment, variance partitioning 6 showed that geographical space alone contributed to a significant proportion of the variation in 7 larval distributions for each species (from 13% to 34% of the total variance). This variance can be 8 attributed to spatial patterns in environmental variables that were not considered here or to 9 ecological processes (Legendre and Fortin, 1989). Key spatial parameters to consider when 10 describing larval distributions are thus the spatially structured characteristics of the species: (1) the 11 location and the size of adult populations that determine potential spawning locations and the 12 amount of released larvae, (2) the characteristics of larval release (e.g., date, intensity, synchrony at 13 the scale of the study area), especially for species with several successive spawning events during 14 an extended reproductive period and (3) the location of the potential sites for settlement. 15 Asynchronous spawning events at the scale of the study area or settlement events can blur the 16 relationships between larval densities and local adult densities. 17 For P. koreni and O. fusiformis, which inhabit subtidal patches of fine sand and muddy fine 18 sand, the locations of adult populations are not exhaustively known and are based on historical data 19 on the distribution of benthic communities (Chassé and Glémarec, 1976). However, annual 20 quantitative samples of benthic macrofauna have been taken since 2004 at one or two sites in the 21 main coastal embayments along the coasts of the northern Bay of Biscay as part of the REBENT 22 monitoring network (www.rebent.org). Although these samples provide only a very crude estimate 23 of stock size, they indicate that densities of both species are highly variable in space at the scale of 24 the surveyed area with large year-to-year fluctuations (Table IV). On average, adult densities of O. 25 fusiformis are higher in the southern part of the study area, in the Vilaine estuary and off Quiberon, 26 and to a lesser extent, in the Bay of Concarneau and in the Bay of Douarnenez, whereas P. koreni is 27 more abundant in the Vilaine estuary and in the Bay of Douarnenez, and to a lesser extent, off 28 Quiberon and in the Bay of Concarneau. A larval distribution more similar to the adult distribution 29 was observed for P. koreni, the study species with the shortest larval planktonic duration (2 weeks), 30 with highest densities reported either off the Vilaine estuary and Concarneau in May, or off the 31 Vilaine estuary, Quiberon and Douarnenez in June. For O. fusiformis with a planktonic larval 32 duration of four weeks, high concentrations of old larvae have been reported in the vicinity of both 33 low-abundance and high-abundance adult populations in May, although our results cannot 34 discriminate between locally-spawned and transported larvae. In contrast, the observation in June of 35 younger larvae in proximity to high-abundance adult populations suggested that local spawning

15 1 heavily influences larval distribution. For both species, temporal variation in larval distribution 2 between the two surveys depended on both intra-seasonal variation in hydrodynamics and the 3 distribution of spawning events. In the case of S. alveolata, the highest larval concentrations 4 observed in the southern part of the study area with a decreasing gradient to the north may be 5 related to the location of the major adult population within the study area in the Bay of Bourgneuf, 6 south of the Loire estuary (Gruet and Lassus, 1983). Interestingly for this species, larval 7 distributions are less governed by geographical space alone than P. koreni and O. fusiformis, while 8 interactions between hydrological environment and geographical space have a greater influence, 9 probably because the spatial distribution of adults partly overlaps the spatial structure of the 10 hydrological environment. 11 Finally, despite the major role played by hydrodynamics, a large proportion of variation in 12 larval distributions, ranging from 14% to 47% was still unexplained and varied among species. Two 13 major processes can contribute to this unexplained portion of the variance: stochastic processes 14 driven by the interaction between coastal circulation and organism life histories (e.g. planktonic 15 larval duration, reproductive events) and biotic factors such as active swimming and vertical 16 migration (Siegel et al., 2008). The stochastic nature of larval dispersal results mainly from the 17 chaotic nature of coastal circulation at the relevant larval time scales and induces patchiness in 18 larval distribution at local spatial scales. Nevertheless, some planktonic larvae can partly regulate 19 their vertical position, either to feed or to avoid predators, in response to environmental cues (e.g. 20 light, tide) or simply because of ontogenic migration. For polychaete larvae, including O. fusiformis 21 larvae, vertical swimming speeds of 1 to 2 mm.s-1 have been measured (Chia et al., 1984; Guizien 22 et al., 2006). These swimming behaviours are thought to greatly influence the horizontal 23 distribution and dispersal of meroplankton through interaction with stratified circulation or tidal 24 currents (Chia et al., 1984; Thiébaut et al., 1992; Hsieh et al., 2010) and create discrepancies 25 between the observed distribution of larvae and the expected distributions for inert particles. For 26 example, in a subtropical estuary, Hsieh et al. (2010) report distributional differences between two 27 groups of polychaete larvae (sabellid larvae vs. spionid larvae) according to their swimming ability. 28 In contrast, if larvae behave as passive particles, their distributions would be expected to be linked 29 to a specific water mass, as has been reported for some and polychaete larvae in Kiel 30 Bay (Banse, 1986) or for some mollusc and polychaete larvae in the estuarine plume of the 31 Chesapeake Bay (Shanks et al., 2002). 32 For the three study species, the analysis of vertical larval distributions revealed contrasting 33 results that can influence their larval horizontal distribution and dispersal and generate species- 34 specific differences. P. koreni and O. fusiformis larvae were mainly distributed above and/or at the 35 thermocline, i.e., within river plume waters, suggesting that their transport may be predominantly

16 1 determined by the displacement of plume waters by surface currents, as previously observed in the 2 Seine river plume (eastern English Channel) (Thiébaut et al., 1992). Nevertheless, the oldest larval 3 stages of these two species were preferentially sampled in bottom layer waters in response to 4 ontogenic vertical migrations. This deeper vertical distribution of old larvae could increase larval 5 retention above adult populations of P. koreni and O. fusiformis in a stratified environment with 6 strong density- or wind-induced currents, such as in estuaries (Lagadeuc, 1992; Thiébaut et al., 7 1992). On the contrary, S. alveolata larvae were evenly distributed over the whole water column 8 even though the different larval developmental stages exhibited a patchy vertical distribution. A 9 previous study in the non-stratified macrotidal environment of the Bay of Mont-Saint-Michel 10 (English Channel) concluded that there is no ontogenic migration for this species, but highlighted 11 short-term variation in relation to the tidal cycle with shallower distribution during flood tides and 12 deeper distribution during ebb tides (Dubois et al., 2007). This type of variation may promote 13 selective tidal-stream transport and therefore inshore larval transport. Moreover, competent S. 14 alveolata larvae have the ability to perceive adult chemical cues to actively select their habitat or 15 delay their metamorphosis if suitable environment for settlement is not encountered (Pawlik, 1988). 16 The influence of vertical migration on larval distribution and dispersal is commonly inferred 17 from variation in the vertical structure of currents or explicitly estimated from biophysical dispersal 18 modelling. However, for weak-swimming larvae such as bivalve or polychaete larvae, this latter 19 approach has given divergent results in different oceanographic environments. In the Chesapeake 20 Bay, North et al. (2008) argue that larval behaviour of two oyster species has greater influence on 21 spatial patterns of larval dispersal than temporal variation in water circulation patterns. In the 22 western Mediterranean Sea, Guizien et al. (2006) highlight that vertical settling of O. fusiformis 23 larvae, which results from a balance between passive settling and active swimming behaviours, is a 24 key parameter in larval dispersal. Conversely, a recent modelling study in the northern Bay of 25 Biscay shows that ontogenic or diel vertical migrations have only a limited impact on the general 26 patterns of larval transport, but decrease the spatial variance of larval dispersal by reducing inter- 27 individual variability in the larval trajectories (Ayata et al., 2010). Based on the present study, 28 disentangling the relative role of stochastic processes and vertical migration on the unexplained 29 variance in larval distribution of the three target species is not possible. However, the fact that the 30 unexplained variance in larval distribution was about twice as high for O. fusiformis than for P. 31 koreni although both species exhibited similar vertical distributions, may suggest a predominant 32 role of stochastic events at local scales. 33 34 CONCLUSION 35 Here, we described the horizontal and vertical distributions of larvae of three coastal polychaetes

17 1 inhabiting patchy subtidal and intertidal environments in relation to the spatial and temporal 2 variability in hydrological mesoscale features recorded in spring in the northern Bay of Biscay. A 3 typological description of the water masses over the continental shelf in spring discriminated high 4 salinity oceanic waters from low salinity river plume waters. The spatial distribution of the 5 hydrological environment was responsible for most of the spatial variation observed in the larval 6 abundances of the three species. Larvae were mainly observed within plume waters, suggesting that 7 the river plume fronts may act as a physical barrier to offshore larval dispersal of coastal 8 invertebrates and favour alongshore transport in the complex and highly variable environment of 9 the northern Bay of Biscay. Cross-shore transport of larvae was mainly governed by wind-induced 10 currents, which influence the location of the river plumes. The spatio-temporal variation in the 11 hydrological environment is thus a key parameter in determining larval dispersal and connectivity 12 and metapopulation dynamics. Our results also underline the role of spatially structured ecological 13 processes, such as the distribution of benthic populations, their size and the timing of spawning or 14 settlement events. Most of the unexplained variability in the larval distributions of each species 15 seemed to be related to stochastic processes in the coastal circulation rather than to larval vertical 16 migrations although ontogenic vertical behaviours were reported for two of the three species (i.e. P. 17 koreni and O. fusiformis). 18 19 ACKNOWLEDGEMENTS 20 We wish to thank the captains and crews of the R/V Côtes de la Manche (INSU-CNRS), as well as 21 R. Michel, V. Ouisse, K. Robert and J. Trigui (Station Biologique de Roscoff) and C. Ellien 22 (University Pierre & Marie Curie) for their valuable help in sampling. Some of the equipment was 23 generously lent by M. Blanchard, P. Gentien and M. Lunven (IFREMER-Brest), J.-C. Brun-Cottan 24 (University of Caen) and É. Grossteffan (IUEM, Brest). We also thank C. Broudin and F. Gentil 25 (Station Biologique de Roscoff) for their data on the benthic macrofauna. This work was supported 26 by the French EC2CO program (Ecosphère continentale et côtière) and by a PhD grant from the 27 French Ministry of Education and Research to SDA. We acknowledge two anonymous reviewers 28 and R. Harris for their constructive comments on the first version of the manuscript. 29 30 REFERENCES 31 Ayata, S.-D., Lazure, P., and Thiébaut, E. (2010) How does the connectivity between populations 32 mediate range limits of marine invertebrates? A case study of larval dispersal between the Bay 33 of Biscay and the English Channel (North-East Atlantic). Prog. Oceanogr., 87, 18--36. 34 Albaina, A. and Irigoien, X. (2007) Fine scale zooplankton distribution in the Bay of Biscay in 35 spring 2004. J. Plankton Res., 29, 851--870.

18 1 Anger, K., Spivak, E., and Luppit, T. (1998) Effects of reduced salinities on development and 2 bioenergetics of early larval stone crab Carcinus maenas. J. Exp. Mar. Biol. Ecol., 220, 287-- 3 304. 4 Banse, K. (1986) Vertical distribution and horizontal transport of planktonic larvae of 5 and benthic polychaetes in an open coastal sea. Bull. Mar. Sci., 39, 162--175. 6 Belgrano, A., Legendre, P., Dewarumez, J., and Frontier, S. (1995) Spatial structure and ecological 7 variations of meroplankton on the French-Belgian coast of the North Sea. Mar. Ecol. Prog. 8 Ser., 128, 43--50. 9 Borcard, D., Legendre, P., and Drapeau, P. (1992) Partialling out the spatial component of 10 ecological variation. Ecology, 73, 1045--1055. 11 Carson, H.S., Lòpez-Duarte, P.C., Rasmussen, L., Wang, D., and Levin, L.A. (2010)!Reproductive

12 timing alters population connectivity in marine metapopulations. Curr. Biol., 20, 1926--1931.! 13 Chassé, C. and Glémarec, M. (1976) Atlas du littoral français : atlas des fonds meubles du plateau 14 continental du golfe de Gascogne : cartes biosédimentaires. Produit numérique REBENT 15 Ifremer-Université-CNRS, 2009 16 (http://www.rebent.org/docs/data/ifr_peupl_ChasseGlemarec_GolfeGascogne_1976_l2_p_EU 17 NIS2004.zip). 18 Chia, F.S., Buckland-Nicks, J., and Young, C.M. (1984) Locomotion of marine invertebrate larvae - 19 a review. Can. J. Zool., 62, 1205--1222. 20 Cowen, R. and Sponaugle, S. (2009) Larval dispersal and marine population connectivity. Annu. 21 Rev. Mar. Sci., 1, 443--446. 22 Dubois, S., Comtet, T., Retière, C., and Thiébaut, E. (2007) Distribution and retention of Sabellaria 23 alveolata larvae (Polychaeta: Sabellariidae) in the Bay of Mont-Saint-Michel, France. Mar. 24 Ecol. Prog. Ser., 346, 243--254. 25 Edwards, K.P., Hare, J.A., Werner, F.E., and Seim, H. (2007) Using 2-dimensional dispersal 26 kernels to identify the dominant influences on larval dispersal on continental shelves. Mar. 27 Ecol. Prog. Ser., 352, 77--87. 28 Fortier, L. and Leggett, W. (1982) Fickian transport and the dispersal of fish larvae in estuaries. 29 Can. J. Fish. Aquat. Sci., 39, 1150--1163. 30 Frontier, S. (1972) Calcul de l’erreur sur un comptage de zooplancton. J. Exp. Mar. Biol. Ecol., 10, 31 121--132. 32 Gawarkiewicz, G., Monismith, S., and Largier, J. (2007) Observing larval transport processes 33 affecting population connectivity progress and challenges. Oceanography, 20, 40--53. 34 Gentil, F., Dauvin, J.-C., and Ménard, F. (1990) Reproductive biology of the polychaete Owenia 35 fusiformis Delle Chiaje in the Bay of Seine (eastern English Channel). J. Exp. Mar. Biol. Ecol.,

19 1 142, 13--23. 2 Gruet, Y. and Lassus, P. (1983) Contribution à l’étude de la biologie reproductive d’une population 3 naturelle de l’annélide polychète Sabellaria alveolata (Linné). Ann. Inst. Océanogr. Paris, 59, 4 127--140. 5 Guizien, K., Brochier, T., Duchêne, J.-C., Koh, B.-S., and Marsaleix, P. (2006) Dispersal of Owenia 6 fusiformis larvae by wind-driven currents: turbulence, swimming behaviour and mortality in a 7 three-dimensional stochastic model. Mar. Ecol. Prog. Ser., 311, 47--66. 8 Hsieh, H.-L., Fan, L.-F., Chen, C.-P., Wu, J.-T. , and Liu, W.-C. (2010) Effects of semidiurnal tidal 9 circulation on the distribution of holo- and meroplankton in a subtropical estuary. J. Plankton 10 Res., 32, 829--841. 11 Irlinger, J.-P., Gentil, F., and Quintino, V. (1991) Reproductive biology of the polychaete 12 Pectinaria koreni (Malmgren) in the Bay of Seine. Ophelia, Suppl 5, 343--350. 13 Jolly, M., Viard, F., Gentil, F., Thiébaut, E., and Jollivet, D. (2006) Comparative phylogeography 14 of two coastal polychaete tubeworms in the Northeast Atlantic supports shared history and 15 vicariant events. Mol. Ecol., 15, 1814--1855. 16 Kelly-Gerreyn, B., Hydes, D., Jégou, A., Lazure, P., Fernand, L., Puillat, I., and Garcia-Soto, C. 17 (2006) Low salinity intrusions in the western English Channel. Cont. Shelf Res., 26, 1241-- 18 1257. 19 Koutsikopoulos, C. and Le Cann, B. (1996) Physical processes and hydrological structures related 20 to the Bay of Biscay anchovy. Sci. Mar., 60, Suppl 2, 9--19. 21 Lagadeuc, Y. (1992) Répartition verticale des larves de Pectinaria koreni en baie de Seine orientale 22 : influence sur le transport et le recrutement. Oceanol. Acta, 15, 109--118. 23 Lagadeuc, Y. and Retière, C. (1993) Critères d’identification rapide des stades de développement 24 des larves de Pectinaria koreni (Malmgren) (annélide polychète) de la Baie de Seine (Manche). 25 Vie Milieu, 43, 217--224. 26 Largier, J. (2003) Considerations in estimating larval dispersal distances from oceanographic data. 27 Ecol. Appl., S13, 71--89. 28 Lazure, P. and Jégou, A.-M. (1998) 3D modelling of seasonal evolution of Loire and Gironde 29 plumes on Biscay Bay continental shelf. Oceanol. Acta, 21, 165--177. 30 Legendre, P. and Fortin, M. (1989) Spatial pattern and ecological analysis. Vegetatio, 80, 107--138. 31 Legendre, P. and Legendre, L. (1998) Numerical Ecology, 2nd English Edition. Elsevier , 32 Amsterdam. 33 Legendre, P. and Trousselier, M. (1988) Aquatic heterotrophic bacteria: modeling in the presence of 34 spatial autocorrelation. Limnol. Oceanogr., 33, 1055--1067. 35 Levin, L. (2006) Recent progress in understanding larval dispersal: new directions and digressions.

20 1 Integr. Comp. Biol., 46, 282--297. 2 Maggs, C.A., Castilho, R., Foltz, D., Henzler, C., Jolly, M.T., Kelly, J., Olsen, J., Perez, K.E., Stam, 3 W., Vainola, R., Viard, F., and Wares, J. (2008) Evaluating signaures of glacial refugia for 4 north atlantic benthic marine taxa. Ecology, 89S, S108--S122. 5 Maguer, J.-F., L’Helguen, S.,Waeles, M., Morin, P., Riso, R., and Caradec, J. (2009) Size- 6 fractionated phytoplankton biomass and nitrogen uptake in response to high nutrient load in the 7 North Biscay Bay in spring. Cont. Shelf Res., 29, 1103--1110. 8 Morin, P., Le Corre, P., Marty, Y., and L’Helguen, S. (1991) Evolution printanière des éléments 9 nutritifs et du phytoplancton sur le plateau continental armoricain (Europe du Nord-Ouest). 10 Oceanol. Acta, 14, 263--279. 11 North, E., Schlag, Z., Hood, R., Li, M., Zhong, L., Gross, T., and Kennedy, V. (2008) Vertical 12 swimming behavior influences the dispersal of simulated oyster larvae in a coupled particle- 13 tracking and hydrodynamic model of Chesapeake Bay. Mar. Ecol. Prog. Ser., 359, 99--115. 14 O’Connor, M., Bruno, J., Gaines, S., Halpern, B., Lester, S., Kinlan, B., and Weiss, J. (2007) 15 Temperature control of larval dispersal and the implications for marine ecology, evolution, and 16 conservation. Proc. Natl. Acad. Sci. USA, 104, 1266--1271. 17 Pawlik, J.R. (1988) Larval settlement and metamorphosis of two gregarious sabellariid polychaetes: 18 Sabellaria alveolata compared with Phragmatopoma californica. J. Mar. Biol. Ass. UK, 68, 19 101--124. 20 Pineda, J., Hare, J., and Sponaugle, S. (2007) Larval transport and dispersal in the coastal ocean and 21 consequences for population connectivity. Oceanography, 20, 22--39. 22 Pingree, R., Pugh, P., Holligan, P., and Forster, G. (1975) Summer phytoplankton blooms and red 23 tides along the tidal fronts in the approaches of the English Channel. Nature, 258, 672--677. 24 Planque, B., Lazure, P., and Jégou, A.-M. (2006) Typology of hydrological structures modelled and 25 observed over the Bay of Biscay shelf. Sci. Mar., 70, 43--50. 26 Puillat, I., Lazure, P., Jégou, A.-M., Lampert, L., and Miller, M. (2006) Mesoscale hydrological 27 variability induced by northwesterly wind of the French continental shelf of the Bay of Biscay. 28 Sci. Mar., 70, 15--26. 29 Queiroga, H., Cruz, T., dos Santos, A., Dubet, J., González-Gordillo, J.I., Paula, J., Peliz, Á., and 30 Santos, A.M.P. (2007) Oceanographic and behavioural processes affecting invertebrate larval 31 dispersal and supply in the western Iberia upwelling ecosystem. Prog. Oceanogr., 74, 174-- 32 191. 33 R Development Core Team (2005). R: A language and environment for statistical computing, 34 reference index version 2.6.2. ISBN 3-900051-07-0, http://www.r-project.org. R Foundation for 35 Statistical Computing, Vienna, Austria.

21 1 Shanks, A., Largier, J., Brink, L., Brubaker, J., and Hooff, R. (2002). Observations on the 2 distribution of meroplankton during a downwelling event and associated intrusion of the 3 Chesapeake Bay estuarine plume. J. Plankton Res., 24, 391--416. 4 Shanks, A., McCulloch, A., and Miller, J. (2003a) Topographically generated fronts, very nearshore 5 oceanography and the distribution of larval invertebrates and holoplankters. J. Plankton Res., 6 25, 1251--1277. 7 Shanks, A., Largier, J., and Brubaker, J. (2003b) Observations on the distribution of meroplankton 8 during an upwelling event. J. Plankton Res., 25, 645--667. 9 Siegel, D.A., Mitarai, S., Costello, C.J., Gaines, S.D., Kendall, B.E., Warner, R.R., and Winters, 10 K.B. (2008) The stochastic nature of larval connectivity among nearshore marine populations. 11 Proc. Natl. Acad. Sci. USA, 105, 8974--8979. 12 Thiébaut, E. (1996) Distribution of Pectinaria koreni larvae (Annelida : Polychaeta) in relation to 13 the Seine river plume front (eastern English Channel). Est. Coast. Shelf Sci., 43, 383--397. 14 Thiébaut, E., Dauvin, J.-C., and Lagadeuc, Y. (1992) Transport of Owenia fusiformis larvae 15 (Annelida: Polychaeta) in the Bay of Seine. I. Vertical distribution in relation to water column 16 stratification and ontogenic vertical migration. Mar. Ecol. Prog. Ser., 80, 29--39. 17 Wilson, D.P. (1932) On the mitraria of Owenia fusiformis Delle Chiaje. Philos. Trans. R. Soc. 18 Lond. B, 221, 231--334. 19 Wilson, D.P. (1968) Some aspects of the development of eggs and larvae of Sabellaria alveolata 20 (L.). J. Mar. Biol. Ass. UK, 48, 367--386. 21 Zarauz, L., Irigoien, X., Urtizberea, A., and Gonzalez, M. (2007) Mapping plankton distribution in 22 the Bay of Biscay during three consecutive spring surveys. Mar Ecol. Prog. Ser., 345, 27--39. 23 Zarauz, L., Irigoien, X., and Fernandes, J.A. (2008) Modelling the influence of abiotic and biotic 24 factors on plankton distribution in the Bay of Biscay, during three consecutive years (2004-06). 25 J. Plankton Res., 30, 857--872. 26 27 28 TABLE LEGENDS 29 Table I: Typology of water masses in the northern Bay of Biscay in May and June 2008. Mean 30 values ± standard deviations of the hydrological variables are indicated for the clusters identified 31 among the sampled stations. Clusters are defined in Fig. 3. 32 33 Table II: Results from multiple regressions on larval abundances and hydrological variables in May 34 2008. The regression coefficients b are given with their associated p-value for each tested factor. 35 The intercept with its associated p-value is indicated in italics and in parentheses. The multiple

22 1 regression was significant for all three species (Pectinaria koreni: R2 = 0.5669, p = 1.863.10-8; 2 Owenia fusiformis: R2 = 0.4486, p = 5.548.10-6; Sabellaria alveolata: R2 = 0.696, p = 3.868.10-12). 3 ***, p<0.0001, **, p<0.001; *, p<0.05. 4 5 Table III: Results from multiple regressions on larval abundances and hydrological variables in 6 June 2008. The regression coefficients b are given with their associated p-value for each tested 7 factor. The intercept with its associated p-value is indicated in italic and in brackets. The multiple 8 regression was significant for all three species (Pectinaria koreni: R2 = 0.6031, p = 2.325.10-9; 9 Owenia fusiformis: R2 = 0.4513, p = 4.945.10-6; Sabellaria alveolata: R2 = 0.7212, p = 4.791.10-13). 10 ***, p<0.0001, **, p<0.001; *, p<0.05. 11 12 Table IV: Adult densities (ind.m-2) of Pectinaria koreni and Owenia fusiformis in different bays 13 along the coasts of southern Brittany. Within each bay, density was estimated as the average of the 14 densities measured at three stations located a few hundred meters apart, and at each station three 15 replicates of 0.1 m2 were collected (C. Broudin and F. Gentil, Station Biologique de Roscoff, 16 unpublished data). na: not available. 17 18 FIGURE LEGENDS 19 Fig. 1: Study area. The locations of the sampled stations are indicated by dark grey dots. The names 20 of the seven transects and the codes of the first and last stations of each transect are indicated. 21 Triangles indicate the locations of the semaphores at Penmarc’h and Belle-Ile where meteorological 22 data were recorded. Light grey areas indicate patches of fine sand and muddy fine sand that form 23 the preferential habitats of P. koreni and O. fusiformis (Chassé and Glémarec, 1976) and where 24 adult populations have been recently reported (Jolly et al., 2006). Black circles indicate locations 25 where adult populations of S. alveolata have been reported (Gruet and Lassus, 1983; F. Rigal and F. 26 Viard, Station Biologique de Roscoff, unpublished data). 27 28 Fig. 2: Surface hydrological conditions in May (A-C) and June (D-F) 2008 in the northern Bay of 29 Biscay: horizontal distribution of sea surface values of (A) salinity in May, (B) temperature (in °C) 30 in May, (C) chlorophyll a concentrations in May, (D) salinity in June, (E) temperature (in °C) in 31 June, (F) chlorophyll a concentrations in June. Sampling transects (with the codes of the first and 32 last stations) are indicated for surface salinity and temperature maps. The Vilaine (V) and the Loire 33 (L) estuaries are indicated on the chlorophyll maps. To minimize cloud coverage, 4-day merged 34 satellite images of surface chlorophyll a concentrations are presented; light-coloured coastal areas 35 indicate higher concentrations (CERSAT data, MORIS/MODIS images). The hydrological maps

23 1 have been performed using ODV (http://odv.awi.de/) with the jetplus color palette proposed by S. 2 Haddock in L&O Bulletin (2010 Vol. 19(2)). 3 4 Fig. 3: Hydrological typology in the northern Bay of Biscay in May (A-B) and June (C-D) 2008: 5 (A) Cluster analysis performed on the hydrological data and (B) spatial distribution of the three 6 clusters according to the surface salinity distribution in May 2008, (C) cluster analysis performed 7 on the hydrological data and (D) spatial distribution of the three clusters according to the surface 8 salinity distribution in June 2008. For each cruise, black squares indicate the stations belonging to 9 cluster 1, grey circles the stations of cluster 2, and white circles the stations of cluster 2'. The 10 hydrological maps have been performed using ODV (http://odv.awi.de/) with the jetplus color 11 palette proposed by S. Haddock in L&O Bulletin (2010 Vol. 19(2)). 12 13 Fig. 4: Horizontal distribution in May (A-C) and June (D-F) 2008 of Pectinaria koreni (A and D), 14 Owenia fusiformis (B and E), and Sabellaria alveolata (C and F) larvae in the northern Bay of 15 Biscay. Percentage of occurrence, mean ± standard deviation of larval abundances calculated over 16 all samples (n = 54), and maximal larval abundance are indicated for each species. 17 18 Fig. 5: Composition of the larval populations: cumulative percentage of the four main larval stages 19 of (A) Pectinaria koreni, (B) Owenia fusiformis, and (C) Sabellaria alveolata in May and June 20 2008 in the northern Bay of Biscay. 21 22 Fig. 6: Variance partitioning of (A) total larval abundances in May, (B) larval stage abundances in 23 May, (C) total larval abundances in June, and (D) larval stage abundances in June. The whole 24 variance of the response matrix (total abundance or larval stages abundances) was partitioned into 25 four fractions a, b, c, and d attributed to the hydrological environment alone, the spatial structure of 26 the hydrological environment, the geographical space alone, and undetermined factors, respectively. 27 Variance partitioning on total larval abundances was performed through partial multiple 28 regressions. Variance partitioning on larval stage abundances was performed through redundancy 29 analyses (see Materials and Methods section). P.k.; Pectinaria koreni; O.f.: Owenia fusiformis; S.a.: 30 Sabellaria alveolata. 31 32 Fig. 7: RDA biplot diagrams of the abundances of the different larval stages of Pectinaria koreni in 33 the northern Bay of Biscay: (A) in May and (B) June 2008. Arrows represent the plot scores of 34 hydrological variables. Sampling sites and larval stages (indicated in grey boxes) are positioned 35 according to their scores along the two axes RDA1 and RDA2. Asterisks indicate significant

24 1 canonical axes (p<0.001). 2 3 Fig. 8: Vertical distribution of the different larval stages of (A) Pectinaria koreni, (B) Owenia 4 fusiformis, and (C) Sabellaria alveolata in the northern Bay of Biscay. Larval abundances per 5 sampling depth are given as larvae m-3. S: surface layer, H/T: halocline and/or thermocline, B: 6 bottom layer. The results are the means obtained for four stations (D2, C1, P2 and V2) sampled in 7 May 2008, and for three stations (D2, P1 and V1) sampled in June 2008. 8 9 SUPPLEMENTARY FIGURE LEGENDS 10 Fig. S1: Hydroclimatic conditions in spring 2008 in the northern Bay of Biscay. (A) Wind 11 conditions at Penmarc’h headland (Météo France data). Daily-averaged zonal (solid line) and 12 meridional (dashed line) wind components are calculated from hourly raw data. Positive zonal and 13 meridional components indicate SW winds. (B) Daily run-offs of the two main rivers of the 14 northern Bay of Biscay from January to September 2008: Vilaine (dashed line) and Loire (solid 15 line). Black arrows indicate the dates of the May and June cruises. 16 17 Fig. S2: Vertical hydrological structures along the transects of Douarnenez, Pouldu-Lorient, and 18 Vilaine in May 2008: (A) temperature, (B) salinity and (C) density profiles. Note that the colour 19 scales differ from those in Fig. 2. 20 21 Fig. S3: Vertical hydrological structures along the transects of Douarnenez, Pouldu-Lorient, and 22 Vilaine in June 2008: (A) temperature, (B) salinity and (C) density profiles. Note that the colour 23 scales differ from those in Fig. 2. 24 25 Fig. S4: Satellite images of the sea surface temperatures in the Bay of Biscay recorded by the 26 NOAA-17 satellite during the two sampling cruises: (A) May 11 20:00 UTC, (B) May 14 20:00 27 UTC, (C) May 18 20:00 UTC, (D) June 09 20:00 UTC, (E) June 11 20:00 UTC, (F) June 18 20:00 28 UTC. Data provided by IFREMER. 29

25 1 TABLE 2 3 Table I 4

Hydrological variables Cluster 1 Clusters 2 and 2' Cluster 2 Cluster 2' Stratification index ! 0.16 ± 0.09 0.03 ± 0.01

Surface temperature Ts 16.20 ± 0.45°C 16.00 ± 0.52°C 14.34 ± 0.71°C

Surface salinity Ss 30.54 ± 1.92 34.39 ± 1.07

Surface density #tS 22.37 ± 1.52 25.58 ± 0.88

Thermocline depth zT 12.22 ± 3.99 m 15.19 ± 4.83 m

Halocline depth zS 10.58 ± 4.32 m 19.94 ± 7.49 m

Pycnocline depth z#$ 10.80 ± 4.19 m 17.64 ± 6.85 m Stratification index ! 0.12 ± 0.04 0.05 ± 0.03

Surface temperature Ts 17.02 ± 1.06°C 17.37 ± 0.86°C 15.59 ± 0.92°C

Surface salinity Ss 31.24 ± 0.67 33.53 ± 0.65

Surface density #tS 22.74 ± 0.38 24.56 ± 0.61

Thermocline depth zT 11.64 ± 3.04 m 14.58 ± 5.57 m

Halocline depth zS 8.07 ± 2.10 m 11.73 ± 5.35 m

Pycnocline depth z#$ 9.01 ± 2.37 m 12.89 ± 5.23 m 5 6

26 1 Table II Species Tested factor Regression coefficient b P-value

TS -0.244 0.036 *

SS -0.283 5.19E-06 ***

P. koreni zS -0.040 0.069 .

z!t 0.009 0.706 (Intercept) (-14.717) (-1.03E-05) ***

TS -0.061 0.619

SS -0.211 8.80E-04 ***

O. fusiformis zS -0.049 0.040 *

z!t 0.023 0.387 (Intercept) (9.279) (0.006) **

TS 0.122 0.177

SS -0.255 4.16E-07 ***

S. alveolata zS -0.044 0.012 *

z!t 0.023 0.242 (Intercept) (7.598) (2.35E-03) ** 2

27 1 Table III

Species Tested factor Regression coefficient b P-value

TS -0.250 4.89E-3 **

SS 0.560 3.32E-3 **

P. koreni z!t -0.047 0.034 * ! 23.626 4.88E-5 *** (Intercept) (-14.287) (0.048) *

TS -0.276 1.24E-4 ***

SS 0.152 0.287

O. fusiformis z!t -0.034 0.049 * ! 6.618 0.116 (Intercept) (0.092) (0.986)

TS -0.150 5.93E-4 ***

SS -0.058 0.506

S. alveolata z!t -0.012 0.261 ! 8.311 2.18E-3 ** (Intercept) (4.432) (0.200) 2 3 4 5 6 7 8 9 10 11

28 1 Table IV 2

Pectinaria koreni 2004 2005 2006 2007 2008 Mean Douarnenez North 3.3 1.1 0 2.2 0 1.3 Douarnenez South na na na 20 2.2 11.1 Audierne 0 0 0 0 0 0 Concarneau 1.1 7.8 11.1 1.1 2.2 4.7 Etel 1.1 2.2 1.1 0 4.4 1.8 Quiberon 11.1 3.3 3.3 0 2.2 4.0 Vilaine Estuary 13.3 25.6 12.2 4.4 4.4 12.0 Vilaine Bay 1 na na na 1.1 1.1 1.1 Vilaine Bay 2 na na na 1.1 0 0.6

Owenia fusiformis 2004 2005 2006 2007 2008 Mean Douarnenez North 17.8 4.4 3.3 4.4 3.3 6.6 Douarnenez South na na na 24.4 23.3 23.9 Audierne 1.1 0 0 0 0 0.2 Concarneau 17.8 16.7 15.6 12.2 4.4 13.3 Etel 3.3 12.2 1.1 0 0 3.3 Quiberon 214.4 61.1 64.4 75.6 78.9 98.9 Vilaine Estuary 57.8 41.1 40 2.2 1100 248.2 Vilaine Bay 1 na na na 1.1 3.3 2.2 Vilaine Bay 2 na na na 106.7 0 53.4 3 4

29